Take the 2-minute tour ×
Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. It's 100% free, no registration required.

CP violation is present in the weak interactions if

  1. There are no degeneracies in the up-quark/down-quark matrices
  2. The Jarlskog invariant $J=Im(V_{us} V_{cb} V_{ub}^* V_{cs}^*)$ is nonvanishing

Furthermore, all CP violating effects are proportional to $J$.

I am getting stuck on showing how all CP violating effects are proportional to $J$. Also, is the Jarlskog invariant a well-known mathematical property of a unitary matrix? What is it quantifying? I'd like to know this to the extent I can generalize this to larger CKM matrices.

Edit: I did a bad job writing my question. I rewrite it here:


  1. How do I constructively derive $J=Im(V_{us} V_{cb} V_{ub}^* V_{cs}^*)$, and how do I generalize this to arbitrary $n\times n$ unitary matrices ?
  2. The Jarlskog invariant is invariant under a change of basis. What is the elegant way of showing this?
share|improve this question

1 Answer 1

Cecilia Jarlskog proposed this invariant already in 1973 and it was mentioned in the original Kobayashi-Maskawa paper.

For three families, it's easy to see why it is nonzero iff the unitary matrix in $U(3)$ can't be brought to the real, orthogonal i.e. $O(3)$ form. It's because after the 5 phase redefinitions of the up-type-quark and down-type-quark eigenstates, every $SU(3)$ matrix may be brought to the form of an $SO(3)$ matrix expressed by 3 real angles $\theta_{ij}$ and a single extra complex phase $\delta$, well, I mean $\exp(i\delta)$, added to a matrix element.

In that parameterization of the $SU(3)$ matrix, the invariant is simply $$ J = c_{12}c_{13}^2 c_{23}s_{12}s_{13}s_{23}\sin \delta. $$ Note that it vanishes exactly if at least one of the factors is zero which means either if the complex phase $\delta$ is zero or $\pi$ mod $2\pi$ – then the matrix is explicitly real orthogonal and CP-preserving – or if any of the sines or cosines of the angles vanish in which case it's also possible to bring the matrix into a real form.



especially pages 7, 8, 11, 12 for some details and formulae. In particular, the first "standard" formula on page 7 makes it clear that the $SU(3)$ matrix is real – or can be made real - whenever one of the factors in $J$ vanishes.


The added questions have nothing whatsoever to do with the original one but they may be answered, too. There is no "constructive way" to derive the Jarlskog invariant. It was a clever guess, a proposed convention. A quantity that is zero whenever it should be is clearly not defined uniquely.

Also, it is incorrect to expect a canonical generalization to larger unitary matrices. Moreover, larger matrices actually have several independent sources of CP-violation, in the same sense as 2 x 2 matrix for 2 families has none. So it would be more natural to have several invariants for larger matrices and say that CP is preserved if all of them are zero. But once again, those invariants wouldn't be unique in any sense.

Concerning the third question, independence on bases, it's trivial to see. The CKM matrix $V$ is the transition matrix mapping three particular mass eigenstates to the $SU(2)$ partners of three other particular eigenstates. All these six eigenstates are determined uniquely, up to a phase (assuming they're normalized).

But it's easy to see that $J$ is invariant under these six changes of phases. For example, change the phase of the $b$ eigenvector by $\exp(i\beta)$. This phase gets cancelled in $J$ because $J$ depends on this phase only via $V_{cb}$ and $V_{ub}^*$ factors: in both of them, $b$ is the second index so the dependence on $\beta$ is the same but the second matrix element is complex conjugated so the phase cancels. Similarly one may verify the cancelation of the five other possible phases and that proves the independence on the basis.

share|improve this answer
Thanks for the answer. But, I already knew most of everything you wrote, which means I asked the wrong question. I have rewritten my question in the edit. Would you please check it again? Thanks! –  QuantumDot Sep 9 '12 at 14:52
Come on, the new questions have nothing to do with the old ones. Jarlskog didn't "derive" it constructively - it was a clever guess, a convention (we're only looking for a quantity that is nonzero when it should be, and the exact value of such a quantity is clearly not unique) and there is no "canonical" generalization of it to larger matrices. For larger matrices, there are many CP-violating angles, not just one, so you should have several invariants, too. –  Luboš Motl Sep 10 '12 at 5:37
Of course my new questions have nothing to do with the original ones-- that's what happens when the original questions weren't the right ones to ask. Ok, now my follow up question to your response is: what is the strategy to follow to construct the different invariants in--say--the 4x4 case –  QuantumDot Sep 10 '12 at 19:43
Sorry, it's a complicated technical question with lots of subtleties of mathematical as well as conceptual nature. I won't be writing a complicated answer, especially because you could end up saying that you were not interested in the answer anyway. –  Luboš Motl Sep 11 '12 at 8:55
Ok, no problem! Typing up a long answer could certainly take a while. Is there a place in the literature where these mathematical subtleties concerning the Jarlskog invariant are discussed? –  QuantumDot Sep 11 '12 at 15:28

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.